THE TRANSPORTATION OF GASES 155 high speeds a given fan will show a marked decrease in efficiency due to slippage, or what with ship propellers is known as cavitation. This, however, is not the case. Methods of Testing Fans. Air-tight-room Test.—In this test the fan discharges into a closed room, the outlet of which is a sliding door or a suitable discharge pipe. The static pressure in the room represents the resistance against which the fan is operating. The sliding door is used to vary the output of the fan, but as the coefficient of efflux varies with different openings as a result of their varying shape, it is better to measure the volume delivered through a discharge pipe of such length as will practically eliminate eddies. The room should be of such size that there are no heavy swirls, and even though extra caution be used in making it tight, a leakage test should be made. With this method of testing it is impossible to get data for a free and unrestricted discharge, yet if the room is relatively large this condition is approached with reasonable closeness. Prof. R. J. Durley1 lets the air discharge from a room or closed vessel through small circular orifices in thin plates of known coefficient of discharge. This, however, is essentially a laboratory method. Discharge Pipe Test.—The discharge pipe, which is preferably of the shape and size of the fan outlet, should have a length of 15 to 20 diameters or 15 to 20 times the mean of the rectilinear dimensions. The Pitot tube should be placed at about 10 diameters from the fan outlet. Traverse readings are taken at this point with a Pitot tube. To get a correct average of the velocity pressure, square-root-of-mean-square-values should be used, but, even with a wide variation in the readings, the inaccuracy of using a plain arithmetical mean is not over 0.5 per cent. Center readings only can be taken, corrected by a coefficient for the pipe if such procedure seems desirable. Credit should be given the fan for the friction loss between the fan outlet and point of measurement. Galvanized iron is the common material of the discharge duct, and the friction loss in inches of water can be computed from the formula H = 4flvz/d, where H = head in feet of air, / = coefficient of friction = 0.0001 (about) for galvanized iron pipe, I = length (ft.), v = velocity (feet per second), and d = diameter of discharge (ft.). This formula may be simplified and written as hf = Ih/kd, where k = a constant, h/ = head lost in friction, and h = velocity head in the pipe, both in inches of water. The value of k for smooth galvanized-iron piping ranges from 50 to 60, while for 12-in. swaged pipe it may be as low as 40 and in some cases is still lower. Readings taken at conical mouthpieces at the end of this discharge pipe are not as accurate as those taken with a Pitot tube in the pipe. Suction Pipe Test.—A similar pipe can be placed on the inlet and Pitot tube readings taken therein. Here the true resistance against which the fan works will be the impact reading. Fan Design.—The matter of fan design will be left to the mechanical engineer and only treated here so far as it will involve choice between different fans. The use of a double discharge, the discharges being 180° apart will decrease both the manometric and volumetric efficiencies. The point where the scroll or spiral discontinues its approach to the circumference of the wheel is called the cutoff point. In some types of fan nearly half the diameter of the wheel is exposed when looking along the axis of discharge, while the other extreme is found where no part of the wheel is exposed, in which case a tangent to the wheel passing through the cutoff point is parallel to the axis of discharge. A fan with a large exposure of wheel will deliver, with very little or no restriction to the flow of air, a relatively greater volume at a given number of revolutions per minute than i Trans. A. S. M. E., Vol. 27.